My key takeaways from the 2026 Lodging Conference – by Chris Haydon, CEO, BirchStreet
Every conference season brings the same promises about artificial intelligence. What stood out to me this year in sunny Phoenix was not the volume of AI talk. Rather, it was the shift in where AI is landing. Profitability pressure set the agenda in many sessions, and the conversations moved from guest-facing experiments toward the back office, where margin is actually made or lost.
The operating layer is fragmented
Ask the people in a hotel group what they need and you hear completely different answers. The property team just wants to run the business. The purchasing group needs every property to stay compliant. The management company focuses on improving profitability. Each stakeholder has a sensible goal, and none of them shares a common layer where those goals meet. Suppliers sit outside that picture entirely.
That gap explains why the most practical advice at the conference was also the least glamorous. Map the handoffs between people and systems. Connect the context before adding intelligence. Let AI earn autonomy in stages: start with rules, add prediction, then recommendations, and only allow automated action where the boundary is clear. Buying more intelligence before your systems agree on basic facts simply makes you wrong faster.
Adoption is moving from the front desk to the back office
Most AI in hospitality is still at the front of the house. Booking, guest messaging, service requests, revenue management and marketing have the most activity, and that makes sense. Those areas are visible and their results are easy to see. But adoption is spreading into operations and finance, and it is moving down from brand technology teams to the owners and management companies who have to prove the effect on the P&L.
That shift changes what success looks like. A brand may measure guest satisfaction scores. An owner measures cost per occupied room, labor hours per shift and the cost of waste. Tools that cannot show movement in those numbers will not survive the next budget review.
Why spend management is a strong place to start
The first is productivity. Every manual purchase order, invoice match and reconciliation takes time away from guests. When routine procurement work is automated, that time returns to the people guests actually see, from housekeeping and food and beverage to front desk staff.
The second is control. Compliance with brand standards, group purchasing agreements and budgets should not depend on someone remembering the rules on a busy Tuesday. Those controls belong inside the transaction itself, so that an off-contract purchase is flagged when it is ordered, not discovered at month end.
An agent will automate whatever disagreement already exists in your data.
Connect the data first
Some operators will stall at this step. Connecting purchasing, receiving, inventory and invoice data is not exciting work, but it is the gate. If the purchase order says one thing, the delivery says another and the invoice says a third, an AI agent will not resolve that conflict. It will process the conflict faster and at greater scale.
Measure what matters to the P&L. Track cost savings, waste reduced and manual handoffs removed. Those measures tell you whether a tool is working, and they give owners a case they can defend.
The data also cannot stay inside the property. Supplier catalogs, pricing history and product content from the wider supply community make adoption faster and insights more actionable. A single hotel has a small sample. A connected network of suppliers and properties has enough context to spot patterns, flag unusual prices and recommend better sourcing.
Final thought
Margin pressure is the mandate for 2027. Start AI where the work is repetitive and the controls matter most, which is spend management. Fix the handoffs and the data before you add intelligence, measure results in P&L terms, and look beyond the four walls of the property. The hotels that win will not be the ones with the most AI. They will be the ones whose data was ready for it.